You are on page 1of 10

How to use the CBI Cluster

In order to run jobs on the Cluster, you must


follow these steps:

1. Connect to the Cluster

2. Determine what type of job you wish to run

3. Write a Script for your job (if needed)

4. Submit the job to be run on the Cluster


Connecting to the Cluster
(From Windows)
1. Download Putty from:
http://the.earth.li/~sgtatham/putty/latest/x86/putty.exe

2. Run Putty

3. In the 'Host Name (or IP address)' text box, type:

<your CBI user name>@cheetah.cbi.utsa.edu

4. Click “Open” at the bottom

5. A window will open. If you are asked if you want to continue


connecting, type “yes” and hit Enter.

6. Enter your CBI password when prompted.


Connecting to the Cluster
(From Mac or Linux)

1. Open a Terminal.

2. Type this command and hit Enter:


> ssh <your CBI username>@cheetah.cbi.utsa.edu

3. If you are asked if you want to continue connecting,


type “yes” and hit Enter.

4. Enter your CBI password when prompted.


After Connecting to the Cluster

cheetah.cbi.utsa.edu is the address of the


cluster's head node, which assigns the jobs to
the other nodes in the cluster.

Please make sure that you DO NOT run the job


directly on this machine because the job will run
on only the head node instead of on the cluster.

The process of correctly submitting jobs to the


cluster will be explained.
Types of Jobs
There are 3 different types of jobs that can be run on the cluster.

Interactive – a job that is run on a single node. If the program you


are running needs interaction from the user to complete, run it
interactively. Interactive jobs should not run for long or be
computationally intensive.

Batch – a job that is run on a single node. These types of jobs are
the same kinds of programs that you would run on your own
computer at home.

Parallel – a job that is run on multiple nodes. The program


executed should be specially designed and compiled to be run
on a cluster. Submitting parallel jobs will be explained in another
guide.
Writing the job Script

It is recommended that you place the job script in


it's own directory, along with all of the files
required by the job. The output file of the job will
be placed in this directory, as well.
Sample Script for a Batch Job

This specifies the name of the job.

This specifies the name of the output file.

#!/bin/bash
#$ -N matlab-test This will cause the error output to be
#$ -o matlabTest.log printed in the output file.
#$ -j y
#$ -cwd This specifies that the job will run in your home directory.
/path/to/program inputFile
exit 0
Submitting Jobs to the Cluster

Batch jobs require a script file before they can be run on the
cluster. Once you have a script file for the job, you can
submit it to the cluster using this command:

> qsub <script file name>

Interactive jobs do not need a script file. To start an


interactive job you can use this command:

> qlogin
Useful Commands for the Cluster

To list all jobs currently running:


> qstat -f

To delete a job that is running:


> qdel <ID of your job>

You might also like